We present a framework for robot skill acquisition, which 1) efficiently scale up data generation of language-labelled robot data and 2) effectively distills this data down into a robust multi-task language-conditioned visuo-motor policy. For (1), we use a large language model (LLM) to guide high-level planning, and sampling-based robot planners (e.g. motion or grasp samplers) for generating diverse and rich manipulation trajectories. To robustify this data-collection process, the LLM also infers a code-snippet for the success condition of each task, simultaneously enabling the data-collection process to detect failure and retry as well as the automatic labeling of trajectories with success/failure. For (2), we extend the diffusion policy single-task behavior-cloning approach to multi-task settings with language conditioning. Finally, we propose a new multi-task benchmark with 18 tasks across five domains to test long-horizon behavior, common-sense reasoning, tool-use, and intuitive physics. We find that our distilled policy successfully learned the robust retrying behavior in its data collection procedure, while improving absolute success rates by 33.2% on average across five domains. Code, data, and additional qualitative results are available on https://www.cs.columbia.edu/~huy/scalingup/.

Large language models (LLMs) such as ChatGPT and GPT-4 have recently demonstrated their remarkable abilities of communicating with human users. In this technical report, we take an initiative to investigate their capacities of playing text games, in which a player has to understand the environment and respond to situations by having dialogues with the game world. Our experiments show that ChatGPT performs competitively compared to all the existing systems but still exhibits a low level of intelligence. Precisely, ChatGPT can not construct the world model by playing the game or even reading the game manual; it may fail to leverage the world knowledge that it already has; it cannot infer the goal of each step as the game progresses.

With the increasing presence of robotic systems and human-robot environments in today's society, understanding the reasoning behind actions taken by a robot is becoming more important. To increase this understanding, users are provided with explanations as to why a specific action was taken. Among other effects, these explanations improve the trust of users in their robotic partners. One option for creating these explanations is an introspection-based approach which can be used in conjunction with reinforcement learning agents to provide probabilities of success. These can in turn be used to reason about the actions taken by the agent in a human-understandable fashion. In this work, this introspection-based approach is developed and evaluated further on the basis of an episodic and a non-episodic robotics simulation task. Furthermore, an additional normalization step to the Q-values is proposed, which enables the usage of the introspection-based approach on negative and comparatively small Q-values. Results obtained show the viability of introspection for episodic robotics tasks and, additionally, that the introspection-based approach can be used to generate explanations for the actions taken in a non-episodic robotics environment as well.

If you use Facebook or YouTube and have seen adverts from web pages you have recently looked at then you have had an experience with artificial intelligence (AI). If you have looked at a recommendation made on a watch-on-demand platform then you have experienced AI. Slowly, AI is being adopted into our lives and if we want to survive as a business we need to learn about AI. AI is a branch of computer science where a machine or programme is capable of doing tasks that generally require human intelligence. Within this branch is machine learning and deep learning.

One of the biggest questions people ask on the internet today is how exactly does artificial intelligence work in email marketing? Yes, they want to know because they have seen how AI is completely reshaping the world of e-commerce, digital marketing, cryptocurrency, real estate, medicine and science in general. You will agree with me that sending emails has been one of the core components of digital marketing. It is asserted that sending mails to the target audience or prospective customers is one of the ways of ensuring that the needed actions are taken. Now that more advanced technologies, such as Artificial Intelligence (AI) have come into play, it is now imperative to explore the advantages and usability of Artificial Intelligence in email marketing.

Cortana may have stopped offering consumer services, but that doesn't mean that Microsoft's virtual assistant is pushing up the virtual daisies in some corner of the metaverse. Instead, she's got a new job, offering a natural language interface into Microsoft 365 services. One of those services is ideal for the new world of hybrid work, where we spend much of our time trying to schedule both physical and online meetings. With meetings needing to be coordinated across internal and external calendars, setting up the average meeting now takes anything up to 30 minutes. Each meeting you're trying to organize adds up to quite a bite out of the workday, a hefty distraction that takes you out of your workflow.

Limited by the time complexity of querying k-hop neighbors in a graph database, most graph algorithms cannot be deployed online and execute millisecond-level inference. This problem dramatically limits the potential of applying graph algorithms in certain areas, such as financial fraud detection. Therefore, we propose Asynchronous Propagation Attention Network, an asynchronous continuous time dynamic graph algorithm for real-time temporal graph embedding. Traditional graph models usually execute two serial operations: first graph computation and then model inference. We decouple model inference and graph computation step so that the heavy graph query operations will not damage the speed of model inference. Extensive experiments demonstrate that the proposed method can achieve competitive performance and 8.7 times inference speed improvement in the meantime.

Text-based games are long puzzles or quests, characterized by a sequence of sparse and potentially deceptive rewards. They provide an ideal platform to develop agents that perceive and act upon the world using a combinatorially sized natural language state-action space. Standard Reinforcement Learning agents are poorly equipped to effectively explore such spaces and often struggle to overcome bottlenecks---states that agents are unable to pass through simply because they do not see the right action sequence enough times to be sufficiently reinforced. We introduce Q*BERT, an agent that learns to build a knowledge graph of the world by answering questions, which leads to greater sample efficiency. To overcome bottlenecks, we further introduce MC!Q*BERT an agent that uses an knowledge-graph-based intrinsic motivation to detect bottlenecks and a novel exploration strategy to efficiently learn a chain of policy modules to overcome them. We present an ablation study and results demonstrating how our method outperforms the current state-of-the-art on nine text games, including the popular game, Zork, where, for the first time, a learning agent gets past the bottleneck where the player is eaten by a Grue.

Since its invention in 1970, email has undergone very little changes. Its ease of use has made it the most common method of business communication, used by 3.7 billion users worldwide. Simultaneously, it has become the most targeted intrusion point for cybercriminals, with devastating outcomes. When initially envisioned, email was built for connectivity. Network communication was in its early days, and merely creating a digital alternative for mailboxes was revolutionary and difficult enough.

Years ago, a mobile app for email launched to immediate fanfare. Simply called Mailbox, its life was woefully cut short -- we'll get to that. Today, its founders are back with their second act: An AI-enabled assistant called Navigator meant to help teams work and communicate more efficiently. With the support of $12 million in Series A funding from CRV, #Angels, Designer Fund, SV Angel, Dropbox's Drew Houston and other angel investors, Aspen, the San Francisco and Seattle-based startup behind Navigator, has quietly been beta testing its tool within 50 organizations across the U.S. "We've had teams and research institutes and churches and academic institutions, places that aren't businesses at all in addition to smaller startups and large four-figure-person organizations using it," Mailbox and Navigator co-founder and chief executive officer Gentry Underwood tells TechCrunch. "Pretty much anywhere you have meetings, there is value for Navigator."